Medicare spent nearly a billion dollars in 2002 on chemotherapy for newly diagnosed elderly breast, colorectal, and lung cancer patients, but surprisingly little is known about the extent to which cancer chemotherapies help or harm such patients. This unsettling paradox is the result of the well-described under- enrollment of the elderly on the clinical trials of chemotherapy. In the absence of trials with representative patients, treating oncologists, patients, and policy-makers are left to extrapolate results of clinical trials conducted in younger and comparatively healthier individuals to the general population with cancer, who tend to be older and have greater comorbid disease burdens. Many basic results of this common extrapolation are unknown. For example oncologists lack even basic observational studies reporting the disease-free survival (DFS) of elderly colorectal (CRC) and breast cancer patients following receipt standard adjuvant chemotherapy. DFS is one of the MOST important end-points in adjuvant chemotherapy in trials, representing the time to cancer relapse or death. Thus, it is relevant to clinicians treating patients in the usual care setting. In this application, our broad aim is to determine whether observational Medicare administrative data may be used to accurately measure DFS in elderly patients with CRC and cancer breast. If it is found to be accurate, the DFS algorithm may meaningful advance research in cancer in the elderly which increasing relies on Medicare data sources. In our work, we study two types of Medicare patients, a highly select cohort of 1,539 elderly breast and CRC patients who were treated with adjuvant chemotherapy in the clinical trial setting and a less select patient cohort of 1,250 elderly patients with the same site and stage-specific cancers and treated with nominally identical adjuvant chemotherapy, but in the usual care setting. By studying these two disparate cohorts, we specifically acknowledge that a valid DFS algorithm that performs well in the clinical trial cohort may not perform well in usual care patients, who form the majority of cancer patients in the United States and who are the focus of a wealth of observational Medicare-based studies of cancer in the elderly. If we find that DFS can be accurately measured in Medicare data, we will provide a critical tool to accelerate research in several fields relevance to cancer in the elderly. Among the fruits of these types of the research will be a clearer understanding of (1) trial efficacy vs. effectiveness which is relevant to both clinical trialists and treating physicians; (2) the clinical effectiveness (CE) of different therapies applied to usual care elderly patients; (3) the patterns of cancer care; (4) the quality of care delivered; and (6) the cost and benefit of specific chemotherapy regimens in the elderly. Regarding the latter, if a therapy does not extend DFS or indeed if it accelerates death, Medicare administrators would need to revisit its reimbursement. Developing an accurate DFS algorithm is a predicate to extending our existing work that examines the risks and benefits of cancer chemotherapy in usual care elderly patients, which touches, to some extent, on all area noted above.